<p>Rock mass quality represents the overall strength and stability of a rock formation. Mapping its spatial distribution is essential for geological evaluations and engineering projects. In mining engineering, resource exploration and continuous excavation yield core samples and surface images, providing substantial geological data for rock mass quality assessment. This study proposes a practical framework for spatial mapping of rock mass quality in mining engineering. A Mask R-CNN deep learning network is developed to automatically calculate RQD from borehole core photos, and a shortest path search algorithm is used to detect discontinuity traces from rock mass exposure images. Geostatistical methods are then used to analyze spatial correlations within the measured rock mass quality data. A heterogenous block model of rock mass quality is established through the spatial interpolation algorithms. Case studies include conveyor belt location optimization in Wushan open-pit mine and differentiated roadway support in Jiama underground mine. The results indicate that the machine vision approaches enable efficient and accurate assessment of rock mass quality. The geostatistics-block method effectively uses measured data to apply spatial interpolation to unmeasured areas. The proposed rock mass quality spatial mapping framework has a broad application potential in mining engineering.</p>